Oct 28, 2026Industry Information
Vision Inspection for Shiny Machined Parts: Questions to Ask Before You Buy
CNC-turned, stamped and bearing parts reflect light back into the camera - why reflective surfaces are hard for vision inspection, and the questions that separate real answers from quotes.
The Part That Breaks the Demo
CNC-turned fittings, stamped terminals, bearing rings and fasteners share one property that makes vision inspection genuinely hard: their surfaces behave like mirrors. Buyers in this space know the pattern - the machine demos beautifully on matte plastic samples, then struggles the moment polished steel goes on the turntable. If your catalog is shiny metal, this article is the checklist to take into the conversation: why reflective surfaces complicate imaging, what no supplier can honestly promise up front, and the questions that reveal whether a supplier has actually handled parts like yours.
Why Reflective Surfaces Are Hard for Cameras
The difficulty is physics, not a defect in anyone's machine. A polished surface does not scatter light back evenly the way a matte surface does - it reflects the lighting and the surrounding environment directly into the lens, so the same part can look bright white in one frame and pitch black in the next, depending on its angle. Curved surfaces (bearing races, ball studs) sweep through that range as they rotate. Add the common variants - oil film from machining, black oxide, anodizing - and the surface can absorb or shift the appearance entirely. Threads and blind holes deepen the problem: their geometry self-shadows, and the feature a buyer cares about (thread quality, a chamfer inside a hole) sits exactly where light reaches worst.
None of this means shiny parts cannot be inspected - metal parts are inspected by vision machines as a normal practice. It means the lighting and imaging setup is part of the engineering, decided per part family - which is why the honest answers to reflective-part questions are demonstrated on samples, not printed in a brochure.
What No Supplier Can Promise Sight-Unseen
One rule protects you in every conversation: repeatability figures (such as ±0.01 mm on the turntable line) describe measurement consistency, not a detection-rate promise - and no detection rate for a reflective part is trustworthy before the supplier has run your actual parts. If a quote declares "99% detection on polished surfaces" without seeing your samples, treat it as a flag, for the same reason a fixed sight-unseen changeover promise (/posts/vision-inspection-machine-changeover) is one. The same rule binds every supplier, us included: what the machine catches on your shiny parts is proven when your parts run, not before. (For scale context, our standard turntable model covers precision hardware from 6 to 80 mm outer diameter, with customization confirmed per product - see the HZXX-JC01 standard model (/posts/hzxx-jc01-single-table-inspection-machine).)
The Reflective-Part Question Checklist
1. What lighting approach do you propose for this geometry? Ask the supplier to describe the setup for your specific part family - angles, diffuse versus directed light - and to show it working on your samples. The answer is engineering, and it should be demonstrated, not asserted. 2. Which of my defects are surface-visible on a shiny part? Take your defect list item by item: a scratch on polished steel, a burr on a thread crest, a missing chamfer inside a blind hole each behave differently. Agree which ones the machine will check and how each will be verified in the sample run. 3. Will you test my worst parts, not my typical ones? Oiled parts straight from the machine, oxidized stock, and the darkest and brightest finishes in your range - the sample test should include your worst case, because reflective behavior is worst-case-driven. 4. How will false rejects be verified? Overkill matters as much as escapes: ask how the sample run separates real defects from acceptable reflections - for example, by running known-NG parts mixed with known-good ones and checking both the reject decisions and the wrongly-rejected rate. 5. What about the dimensional checks? ID, OD and cross-section measurements on reflective edges carry the same question: how is the edge located consistently on a shiny part, and what repeatability was confirmed on similar parts? 6. How does a multi-SKU shop handle changeover? Reflective metal usually means many part numbers; the changeover questions (what physically changes between families, how setups are stored) are covered in the changeover guide above. 7. What records come out of the run? For customer complaints and audits you will want the images and decisions logged, as described in vision inspection data traceability (/posts/vision-inspection-data-traceability).
What to Send for a Real Answer
The sample package decides everything here. Send the parts that behave worst: the oiliest, the most polished, the most oxidized, the ones your inspectors argue about. Include the defect list with a marked example of each defect, and the dimensional tolerances for the features that matter. Sample quantity, logistics and confidentiality terms are confirmed when you inquire. The sample run returns inspection images and the decision log - which is precisely the evidence a reflective-part decision should rest on. Inputs that shape the inquiry are listed in the quote preparation checklist (/posts/vision-inspection-machine-quote-checklist).
Frequently Asked Questions
Can vision inspection handle polished steel at all?
Reflective metal parts are inspected by vision machines as a normal practice; what varies is the lighting and imaging setup per part family. Whether a specific machine handles your specific finish is demonstrated on your samples - that demonstration, not a brochure claim, is the answer.
What lighting works on shiny metal surfaces?
As general imaging practice, diffuse or dome lighting softens reflections by scattering light from many angles, while tightly directed light tends to fail on mirror-like surfaces. The actual setup for your part - angles, intensity, configuration - is engineering that gets demonstrated on your samples.
Why did the demo look great but our parts test badly?
Demo parts are usually chosen to show well - matte finishes, flat features. Your production parts carry the reflective behavior, oils and geometry that stress the imaging. Always demand the sample test on your own parts, including the worst ones.
Are threads and blind holes checkable?
They are checkable where the geometry allows the imaging to reach the feature, and the result is confirmed on samples with marked defect examples. Threads and deep blind holes are exactly where the "which defects are surface-visible" question above earns its place.
Do you guarantee a detection rate for our shiny parts?
No - and be careful with any supplier who does sight-unseen. Detection on your parts is confirmed in the sample test; the brochure figures that do exist (repeatability, throughput ranges with their conditions) describe the machine, not an outcome on your defects.
Next Step
Send your worst-case metal parts with the marked defect list and tolerances. We run the sample test and return the inspection images and the decision log - the only basis on which a reflective-part decision should be made. Contact the team (/contact-us) - Monica (+86 138 5726 5705) or Arya (+86 173 2430 9735), or email monica@aivisioncheck.com.
